"Vibe Coding" - lovable? It's simple, but I'm not sure I see the benefit

This feels like "cheating" at prompting - but I've yet to understand what it offers over and above creating tools in an LLM itself. Maybe I should have pushed it further.

Having spent quite a bit of time learning to get good results from prompts in Large Language Models I was introduced recently to “Vibe Coding” - using a platform that creates a result from simpler prompts and coverts directly into code.

This weekend the platform Loveable had an offer on, making all projects free so I thought I’d give it a try.

This tool (for reviewing contracts and producing a red/amber/green review and suggestions) took all of 20 minutes to create. It is simple, has no context, no legal direction from me, no examples and I do not know where the data is stored (so must not be used for anything confidential or containing personal data)

But if you’re signing up to generic SAAS terms and conditions and want a quick view as to what to question, then this might be a very rough starting point. As to what to change and how - you likely need a lawyer for that.

Note - it didn’t require any legal knowledge to build it an is a curiosity, definitely not reliable nor legal advice.

I haven’t really worked out what the advantage is over creating a GPT in OpenAI itself (or a Gem in Gemini or using co-pilot Studio), but perhaps that is because this is a simple tool and there are variations that I could make, new functionality perhaps that would make vibe coding different to creating something in the LLM itself.

I’d be interested in knowing what other people have found. Is there an advantage to Vibe Coding compared to simply creating tools in the LLM itself? (perhaps that the user isn’t required to have an account with the same LLM?)